HaProxy container stopped working after a few days

openBalena
1

Hi,

I just found a problem with the haproxy container. It stopped working for an unknown reason after being used for several days without issues. It’s actually the second time this issue happens.

Restarting the haproxy container actually fixes the issues, so maybe a simple ‘restart on failure’ strategy would be enough ?

2019-04-23T15:47:56.350857362Z [NOTICE] 112/154756 (15) : New worker #1 (17) forked
2019-05-19T04:28:27.563161482Z [ALERT] 138/042827 (15) : Current worker #1 (17) exited with code 139 (Segmentation fault)
2019-05-19T04:28:27.587340561Z [ALERT] 138/042827 (15) : exit-on-failure: killing every workers with SIGTERM
2019-05-19T04:28:27.587371941Z [WARNING] 138/042827 (15) : All workers exited. Exiting... (139)

Thanks for your help, love your work !

3

Thanks for this. I will take a look and discuss internally; it would make sense to understand why HAproxy is dying but the restart-on-failure is a good option either way, in my view.

1 Like
5

Happened again today around noon…

6

I’d suggest doing some checking on resources in use prior to crashes, CPU, Memory, and Network - see what is going on.

HAProxy will crash under extremely high load - which can happen quite easily if its deployed with publicly available endpoints.